This study tackles unsupervised domain adaptation of reading comprehension (UDARC). Reading comprehension (RC) is a task to learn the capability for question answering with textual sources. State-of-the-art models on RC still do not have general linguistic intelligence; i.e., their accuracy worsens for out-domain datasets that are not used in the training. We hypothesize that this discrepancy is caused by a lack of the language modeling (LM) capability for the out-domain. The UDARC task allows models to use supervised RC training data in the source domain and only unlabeled passages in the target domain. To solve the UDARC problem, we provide two domain adaptation models. The first one learns the out-domain LM and in-domain RC task sequentially. The second one is the proposed model that uses a multi-task learning approach of LM and RC. The models can retain both the RC capability acquired from the supervised data in the source domain and the LM capability from the unlabeled data in the target domain. We evaluated the models on UDARC with five datasets in different domains. The models outperformed the model without domain adaptation. In particular, the proposed model yielded an improvement of 4.3/4.2 points in EM/F1 in an unseen biomedical domain.

Kosuke Nishida, Kyosuke Nishida, Masaaki Nagata, Atsushi Otsuka, Itsumi Saito, Hisako Asano, Junji Tomita. Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. 2019.

Kyosuke Nishida, Itsumi Saito, Kosuke Nishida, Kazutoshi Shinoda, Atsushi Otsuka, Hisako Asano, Junji Tomita. Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. 2019.

This study considers the task of machine reading at scale (MRS) wherein, given a question, a system first performs the information retrieval (IR) task of finding relevant passages in a knowledge source and then carries out the reading comprehension (RC) task of extracting an answer span from the passages. Previous MRS studies, in which the IR component was trained without considering answer spans, struggled to accurately find a small number of relevant passages from a large set of passages. In this paper, we propose a simple and effective approach that incorporates the IR and RC tasks by using supervised multi-task learning in order that the IR component can be trained by considering answer spans. Experimental results on the standard benchmark, answering SQuAD questions using the full Wikipedia as the knowledge source, showed that our model achieved state-of-the-art performance. Moreover, we thoroughly evaluated the individual contributions of our model components with our new Japanese dataset and SQuAD. The results showed significant improvements in the IR task and provided a new perspective on IR for RC: it is effective to teach which part of the passage answers the question rather than to give only a relevance score to the whole passage.

This study focuses on acquisition of commonsense knowledge.A previous study proposed a commonsense knowledge base completion (CKB completion) method that predicts a confidence score of triplet-style knowledge for improving the coverage of CKBs.To improve the accuracy of CKB completion and expand the size of CKBs, we formulate a new commonsense knowledge base generation task (CKB generation) and propose a joint learning method that incorporates both CKB completion and CKB generation.Experimental results show that the joint learning method improved completion accuracy and the generation model created reasonable knowledge.Our generation model could also be used to augment data and improve the accuracy of completion.

Modern Japanese texts often include Western sourced words written in the Roman alphabet. Even Japanese sourced words are sometimes transliterated into the Roman alphabet. As most of them are very new and idiosyncratic proper nouns, it is impractical to assume all those alphabetic words can be registered in the word dictionary of a text-to-speech system; their pronunciation must be derived automatically. As long vowel expressions are the same with short vowel or diphthong expressions in Japanese words transliterated into the alphabet, long vowel detection is necessary for generating highly accurate pronunciation. This paper proposes a method to detect long vowel positions using the Support Vector Machine. The cost of making the learning model for the detection is very low because the training data can be generated automatically from the text-to-speech word dictionary. 93.2% word accuracy was achieved which is almost the maximum accuracy possible from just spelling information.

This paper shows by a quantitative evaluation that nonrestrictive modication in Japanese is a marked factor of prosody, and describes how to extract non-restrictive modication.In Japanese, restrictive and non-restrictive modication have the same syntactic structure, so it is dicult to distinguish them.We propose rules to extract non-restrictive modication in news-text based on the classication of nonrestrictive modication and the combination patterns of the head noun and the modier.